128 I The Process of Evolution 



pelling is the strong correlation, in time and space, of industrializa- 

 tion and the appearance of the melanic forms. This is so striking as 

 to make virtually certain some relationship between the two phe- 

 nomena. The composition of various English populations of Biston 

 hetularia is shown in Fig. 7.3. The populations in the industrial mid- 

 lands are highly melanic, as are those in eastern England on the 

 downwind side of the industrial areas where pollution fallout has 

 been most intense. 



The most elegant demonstration of one factor responsible for in- 

 dustrial melanism is the observations and experiments on differential 

 predation by Kettlewell and Tinbergen. In an unpolluted wood in 

 Dorset, equal numbers of melanic and typical individuals were re- 

 leased; predation was observed and photographed from a blind. 

 Spotted flycatchers (Muscicapa striata), nuthatches {Sitta euro- 

 paea), yellow hammers (Emberiza citrinella), robins (Erithacus 

 rubecuh), and thrushes (Tiirdus ericetonim) ate 164 melanic in- 

 dividuals but only 26 typical individuals (P <^ .01).Mn a polluted 

 wood near Birmingham, opposite results were recorded, with red- 

 starts (Phoenicurtis phoenicurus) eating only 15 melanics in contrast 

 to 43 typical individuals (P <^ .01). Very impressive motion pictures 

 were made of these experiments, with repeated sequences of birds 

 searching tree trunks and eating the conspicuous moths without 

 noticing adjacent protectively colored individuals. 



A series of release and recapture experiments supported the 

 visual-predation hypothesis. Known numbers of marked individuals 

 of both types were released in an area, and the percentage of each 

 recaptured later at a light was recorded. The percentage of recov- 

 ered moths contrasting with their background (melanics in un- 

 polluted and typicals in polluted areas) was considerably lower 

 than the percentage of cryptically colored moths recaptured. For 

 instance, near Birmingham (where the population is 85 to 87 per- 

 cent melanic) 154 melanics and 73 typicals were marked and re- 

 leased. Later 98 marked moths were recaptured, 82 melanics (53 

 percent of 154 released ) and 16 typicals ( 25 percent of 64 released ) . 

 If experimental error is ignored and the survival value of the favored 

 melanics is set equal to 1, the survival value (1 — s) of the typical 



' Probaliility much less than 1 in 100. This means that a statistical test has 

 indicated that a deviation from the expected frequencies of this magnitude 

 would, on the basis of chance alone, be expected much less than 1 percent of 

 the time. The notation (P < .02) means probability less than 2 in 100. Many 

 biologists conventionally consider "significant" any result where the probability 

 of chance alone being responsible is less than 5 percent. An elementary discus- 

 sion of the use of statistical inference in biology may be found in Simpson, Roe, 

 and Lewontin, Quantitative Zoology. 2d ed. Harcourt, Brace & Co., New York. 



